Background
Cell-free fetal DNA analysis is used as a screening test to identify pregnancies that are at risk for common autosomal and sex chromosome aneuploidies.
Objective
The purpose of this study was to investigate the chromosomal abnormalities that would not be detected by cell-free fetal DNA in a single medical center.
Study Design
This was a retrospective cohort analysis of 3182 consecutive invasive diagnostic procedures that were performed at Montefiore Medical Center’s Division of Reproductive and Medical Genetics from January 1, 2009 to August 31, 2014. All patients underwent cytogenetic analysis; one-third of the patients (1037/3182) went through chromosomal microarray analysis.
Results
Clinically significant chromosomal abnormalities were detected in 220 of 3140 cases (7%) after we excluded multiple gestation pregnancies (n = 42). Of these 125 cases (57%) were diagnosed with the common autosomal trisomies that involved chromosomes 21, 18, and 13 and with sex chromosome aneuploidies. There were 23 mosaic karyotypes; 8 of them involved trisomy in chromosomes 21 and 13; 5 of them were sex chromosome mosaics, and 10 of them were other mosaic cases. Five cases of triploidy were detected. Additionally, 19 unbalanced chromosomal rearrangements, a rare autosomal trisomy, and 47 clinically significant findings on chromosomal microarray analysis were diagnosed. Based on the published detection rates of cell-free fetal DNA testing and considering the “no-results” rate, we calculated that 99 of 220 chromosomal changes (45%) could not have been detected by cell-free fetal DNA testing: 16 of the 125 common aneuploidies and sex chromosome aneuploidies, 1 of the 5 triploidy cases, 15 of the 23 mosaic cases, all cases of unbalanced chromosomal rearrangements (n = 19), rare autosomal trisomy (n = 1), and 47 clinically significant chromosomal microarray abnormalities.
Conclusions
Current cell-free DNA testing could not detect up to one-half of the clinically significant chromosomal abnormalities that were found, which included clinically significant chromosomal microarray abnormalities. Among the 99 abnormal karyotypes that were not identified by cell-free DNA screening, 79% were from women with abnormal screening or abnormal ultrasound finding; 21% were from women who underwent invasive testing simply for advanced maternal age/concern, with no other risk factors or ultrasound findings. This information highlights the limitations of cell-free DNA screening and the importance of counseling patients about all prenatal screening and diagnostic procedures and about the added gain of invasive testing with karyotype and microarray.
Cell-free fetal DNA (cffDNA) testing is a screening test that shows unsurpassed sensitivity for the detection of trisomy 21, both in the high-risk and the low-risk population. CffDNA testing also shows good results in the identification of pregnancies that are at risk for other common autosomal aneuploidies (trisomy 18 and trisomy 13). Detection of sex chromosome abnormalities is also improving, and recent studies have shown promising results. Detection rates (DRs) for mosaics currently are undetermined, and the detection of triploidy with cffDNA depends on the method that is used and on whether it is a diandric or a digynic triploidy. Many national organizations have set guidelines for the use of cffDNA for aneuploidy screening with a collective conclusion that patients who are at increased risk for aneuploidy can be offered cffDNA screening with appropriate pretest counseling.
Amniocentesis and chorionic villus sampling (CVS) are invasive diagnostic procedures for the investigation of fetal chromosomal and subchromosomal abnormalities; both carry a risk for miscarriage. According to the recent metaanalysis, the weighted pooled procedure-related risks of miscarriage for amniocentesis and CVS were 0.11% (95% CI, –0.04 to 0.26%) and 0.22% (95% CI, –0.71 to 1.16%), respectively.
The National Institutes of Health (NIH)–sponsored clinical trial investigated the accuracy of fetal diagnosis by comparing metaphase karyotype and chromosomal microarray analysis (CMA) and showed that there is an increase in the detection of clinically significant CMA abnormalities, even when the metaphase karyotype was normal.
It was reported previously that 17.4% of pregnancies with a positive quadruple test result had karyotype other than the common trisomies (trisomy 21 or trisomy 18/13). Additionally, among patients who underwent invasive prenatal diagnosis because of a positive first-trimester screening (FTS), nearly 30% of the patients were found to have a chromosomal abnormality on karyotype other than the common trisomies ; however, CMA abnormalities were not included in these studies.
It is of concern that the use of cffDNA to rule out trisomies 21,18, or 13 after a positive first- or second-trimester screening test might result in a diminution in the chromosomal abnormalities (microscopic and submicroscopic) that can be detected with the use of the current invasive procedures.
We aimed to ascertain the percentage of chromosomal abnormalities that would be missed if only cffDNA testing was performed in an underserved, high-risk population.
Materials and Methods
We report a retrospective cohort analysis of 3182 consecutive amniocentesis and CVS procedures performed at Montefiore Medical Center’s Division of Reproductive Genetics from January 1, 2009 (the introduction of CMA in our center) to July 31, 2014.
All women who are treated at our medical center are offered a traditional screening test: FTS (nuchal translucency [NT] and analytes) when they initiate prenatal care early in pregnancy or quadruple screening (analytes alone) for patients who need prenatal care past the first trimester and up to 21 weeks gestation. A mid-trimester detailed anatomy scan is offered to all patients. High-risk patients and the patients who are interested in invasive testing are referred for genetic counseling. If a patient chooses to have a diagnostic test, a CVS or amniocentesis (10 to 13 + 6/7 and 16-23 weeks gestation, respectively) is performed. Invasive procedures are performed on-site at our center; a standard metaphase cytogenetic analysis of cells that are obtained by amniocentesis or CVS is performed in one of the authorized diagnostic laboratories routinely used by our institute. Array-based comparative genomic hybridization (aCGH) has been used at our center since 2009 (2009-2010 as part of an NIH array study that used mainly oligonucleotide probes and since 2010 have used a single nucleotide polymorphism [SNP] platform for most patients and oligonucleotide platform for the remainder of patients, depending on insurance coverage and referent laboratory). Until 2013, aCGH was offered only to high-risk patients who were having an invasive procedure (high risk includes advanced maternal age [AMA] and maternal age adjusted risk after screen positive test, women who had a previous fetus/child affected by autosomal trisomy, structural anomalies identified by ultrasonography and parental carrier of chromosomal rearrangement ). Beginning in 2014, aCGH was offered to all patients who would undergo an invasive procedure as per the American College of Obstetricians and Gynecologist recommendation.
All results were recorded in the patient’s electronic medical record and the department’s log books by board-certified genetic counselors and were reviewed by medical geneticists. Results were categorized into common aneuploidies (involving trisomies in chromosomes 21, 18, and 13), sex chromosome aneuploidies (monosomy X, XXX, Klinefelter syndrome and XYY syndrome), triploidy, unbalanced chromosomal rearrangements (translocation, inversion and deletion/duplication), mosaics, and CMA abnormalities.
Statistical analysis
The Student t test and Pearson’s chi-square test were used to evaluate the statistical significance of the comparison of the indication for procedure in the normal and abnormal results groups. A probability value of <.05 was considered to indicate statistical significance.
Calculation of detectability by cffDNA
We used the following weighted pooled DR and false-positive rates (FPR), based on the recent metaanalysis of studies of maternal peripheral blood cffDNA analysis : For trisomy 21, 99.2% DR (95% CI, 98.5–99.6%) with 0.09% FPR (95% CI, 0.05–0.14%); for trisomy 18, 96.3% DR (95% CI, 94.3–97.9%) with 0.13% FPR (95% CI, 0.07–0.20%); for trisomy 13, 91.0% DR (95% CI, 85.0–95.6%) with 0.13% FPR (95% CI, 0.05–0.26%); for monosomy X, 90.3% DR (95% CI, 85.7–94.2%) with 0.23% FPR (95% CI, 0.14–0.34%); for sex chromosome aneuploidies other than monosomy X, 93.0% DR (95% CI, 85.8–97.8%) with 0.14% FPR (95% CI, 0.06–0.24%).
The DR of triploidy was calculated based on available publications at this time; aiming to identify fetal triploidy using cffDNA. Nicolaides et al showed the correct identification of 4 of 4 diandric triploidy using a SNP-based cffDNA. That same method failed to detect 4 of 4 digynic triploidy. Others also demonstrated very low fetal fraction in digynic triploidy (fetal fraction <3%, no result reported on cffDNA) and correct identification of diandric triploidy with the use of SNP. Hence, a 100% DR for diandric triploidy with SNP-based cffDNA testing and 0% DR for digynic triploidy were assumed.
Clinical validation trials report a wide range of DRs of mosaics cases by cffDNA analysis ; 3 of 3 of mosaic trisomy 21 and 1 of 1 mosaic trisomy 18 were detected by cffDNA; the DR reported for monosomy X mosaic is 2 of 7 (29%) . CffDNA test has not been reported to detect other complex mosaics. Others have reported cffDNA analysis to detect only 1 of 2 cases of mosaic trisomy 21. Also, cffDNA could not detect mosaic trisomy 13 and mosaic trisomy 21 superimposed with mosaic T18 (trisomy 21 was detected, but the mosaic T18 was not). We conservatively calculated the common trisomies mosaic DR as the same as we calculated for the complete aneuploidies. This is probably an overestimation because the contribution of the fetal excess chromosome is partial; therefore, the DR is expected to be lower compared with the DR of the complete trisomies. For monosomy X mosaic, we calculated 29% DR. For other sex chromosome aneuploidy mosaic, there is no published DR.
To date, there are no data available for the cffDNA DR of unbalanced chromosomal rearrangements or CMA abnormalities.
Data on the no-result rate because of assay failure or low fetal fraction varies dramatically from 0.5 2 -6.1% and recently 3% in the United States and 0.1% in China (with 2.18% required repeat blood sampling). In pregnancies that are complicated with chromosomal aneuploidies, the rate of no-result is increased. Pergament et al reported 16% of aneuploidies had no-result on cffDNA analysis. On the basis of the recent metaanalysis by Gil et al, we assumed that, in 6.9% of trisomies (complete and mosaic) and in 17.2% of sex chromosome aneuploidies (complete and mosaic), a no-result will be received.
The study was approved by the Albert Einstein Medical Center institutional review board (IRB Number: 2014-4252).
Results
Between January 1, 2009, and July 31, 2014, 3182 consecutive procedures were performed at Montefiore Medical Center’s Division of Reproductive Genetics: 2514 amniocenteses and 668 CVSs. Forty-two procedures were excluded because of multiple gestations: 33 procedures were done for 20 twin pairs (of them 6 were monochorionic) and 9 procedures were done for 2 triplet and 1 quadruplet pregnancies.
All 3140 samples had standard karyotype; 1037 samples also had aCGH performed. Maternal age ranged from 15-55 years (median, 38 years). Indications for the procedure in the study cohort divided by normal and abnormal results groups are listed in Table 1 .
Indication | Normal results group a (n = 2920), n (%) | Clinically significant abnormal result group (n = 220), n (%) | P value |
---|---|---|---|
Advanced maternal age | 858 (29.4) | 30 (13.6) | <.0001 |
Increased nuchal translucency | 144 (4.9) | 47 (21.4) | <.0001 |
Abnormal ultrasound finding | 501 (17.2) | 67 (30.1) | <.0001 |
Known parental chromosomal rearrangement carrier status | 219 (7.5) | 4 (1.8) | <.01 |
Previous effected pregnancy | 105 (3.6) | 3 (1.4) | .08 |
Maternal concern | 169 (5.8) | 2 (0.9) | <.01 |
Elevated alpha fetoprotein on second trimester screen or suspected neural tube defect | 95 (3.2) | 3 (1.4) | .12 |
Abnormal first trimester screen | 333 (11.4) | 33 (15) | .16 |
Abnormal quadruple screen | 452 (15.5) | 21 (9.5) | <.05 |
Abnormal cell-free fetal DNA results | 4 (0.1) | 8 (3.5) | <.0001 |
Other | 33 (1.1) | 2 (0.9) | |
Isoimmunization | 7 (0.2) | 0 | .46 |
Chromosomal abnormalities were determined by karyotype and aCGH. Of the 3140 standard metaphase cytogenetic analyses, chromosomal changes were detected on 208 karyotypes. Of them, there were 97 cases of common autosomal aneuploidies (involving chromosomes 21, 18, 13), 1 case of trisomy 16 on CVS, 28 cases of sex chromosome aneuploidies (21 monosomy X and 7 sex chromosome trisomies), 19 unbalanced rearrangements, and 5 triploidy cases. Twenty-three mosaic karyotypes were detected: 6 cases of trisomy 21, 2 cases of trisomy 13, 5 cases that involved sex chromosomes, and 10 other rare complex mosaic cases ( Table 2 ). Of note, the CVS for the case of trisomy 16 was performed because of positive FTS. The patient did not have CMA; however, the fetus had complex cardiac and structural anomalies.
Abnormalities on karyotype and chromosomal microarray analysis (n = 249) | Detected, n | Cell-free fetal DNA | ||
---|---|---|---|---|
Detection rate, % | Failure rate, % | Calculated detection rate in the cohort | ||
Any autosomal aneuploidy | 98 | |||
Trisomy 21 | 61 | 99.2 | 6.9 | 61X0.992X0.931=56.3 |
Trisomy 18 | 32 | 96.3 | 6.9 | 32X0.963X0.931=28.7 |
Trisomy 13 | 4 | 91.0 | 6.9 | 4X0.91X0.931=3.4 |
Other autosomal trisomy (trisomy 16) | 1 | NA | 0/1 | |
Any sex chromosome aneuploidy | 28 | |||
45X | 21 | 90.3 | 17.2 | 21X0.903X0.828=15.7 |
47XXX/47XXY/47XYY | 1/3/3, respectively | 93.0 | 17.2 | 7X0.93X0.828=5.4 |
Any structural rearrangement | 43 | |||
Balanced | 24 | NA | Not investigated in our cohort | |
Unbalanced | 19 | NA | 0/19 | |
Triploidy | 5 | 100, Diandric using single nucleotide polymorphism technology | 0, Diandric and 100 digynic triploidy | 4/5 |
All mosaic | 28 | |||
Common autosomal aneuploidy mosaic | 8 | 99.2 a | 6.9 | 8X0.992X0.931=7.4 |
Other autosomal aneuploidy mosaic | 4 | NA | NA | 0/4 |
Sex chromosome aneuploidy mosaic | 2 monosomy X, 3 others | 29 3 NA | 17.2 NA | 2X0.29X0.828=0.5 0/3 |
Other complex mosaics | 6 | NA | NA | 0/6 |
Confined placental mosaicism b (involving chromosomes 2, 10, 16, 11, 20) | 5 | NA | NA | Not investigated in our cohort |
Submicroscopic chromosomal abnormalities on chromosomal microarray analysis | 47 | NA | NA | 0/47 |
a A conservative estimation based on the detection rate published for complete trisomies, because of the lack of consensus regarding the detection rate of mosaics with the use of cell-free fetal DNA analysis
b Includes 1 case of low level mosaicism on chromosomal microarray analysis that was performed on cultural cells; the patient declined amniocentesis, and no confirmation was done.
Thirty-five cases were excluded from the 208 abnormal karyotype findings; 24 cases were parentally inherited balanced chromosomal rearrangements; another 6 results were duplicated; they were first noted on CVS (included in the study cohort) and later confirmed on amniocentesis (not included in the study). In addition, another 5 results were considered confined placental mosaics. All confined placental mosaics were tested for uniparental disomy (UPD). Thus, 173 of our 3140 cases (5.5%) were considered to have clinically significant chromosomal abnormalities that were found on karyotype ( Figure 1 ).
Array CGH testing was performed in 1037 fetuses during the research period (33% of the patients). Over one-third of the aCGH tests (401/1037; 38.7%) were performed for fetal anomalies that were detected on ultrasonography (first-trimester NT and anatomy scan). Of the 1037 CMA studies, 870 aCGH results were considered normal. There were 100 abnormal CMA results; 53 of 100 reflected the abnormal karyotype findings, and 47 of 100 were clinically significant CMA abnormalities in the presence of normal karyotype (20 cases were considered and counseled as pathogenic changes; 23 cases were considered and counseled as likely pathogenic: 2 cases of loss of heterozygosity [LOH] pathogenic and 2 cases of UPD-likely pathogenic). Forty-eight cases of variants of unknown significance were considered benign variants and thus were excluded. Additionally there were 16 cases of nonclinically significant LOH and 1 case of nonclinically significant UPD that were excluded. In 2 cases, the laboratory failed to report the CMA results ( Figure 2 ).
Microarray tests were considered clinically significant if an Online Mendelian Inheritance in Man (OMIM)–annotated gene was recognized in the deleted/duplicated interval and/or the deletion in the region is known to have a clinical significance. LOH was considered clinically significant if it involved known OMIM genes or if ultrasound scan revealed major anomalies that could be attributed to the LOH. UPD was considered clinically significant if previously described in the literature as having potential deleterious outcome.
A total of 220 clinically significant chromosomal abnormalities (7%; 173 microscopic and 47 submicroscopic) were detected in our population (individual karyotype and CMA findings listed in the Supplemental Table ).
There were 16 cases of failed results; 7 on amniocentesis (0.28%), 9 on CVS (1.3%), and 2 CMA failures (0.2%)
Table 2 summarizes the calculated predicted cffDNA DR in our cohort, based on the published weighed pooled DRs and no-result rates of cffDNA screening.
For the common trisomies and sex chromosome aneuploidies, we calculated that 109 of 125 cases (87.5%) would be detected by cffDNA analysis. Because of lack of consensus regarding the common trisomy mosaicism, we assumed conservatively that the DR and the no-result rates for these mosaics are the same as for the common trisomies. We predicted that 7 of 8 cases of mosaic trisomy 21 and 13 would have been detected by cffDNA. In addition, we predicted that 1 of 2 mosaic monosomy X cases would have been detected by cffDNA. A total of 8 of 23 cases of mosaics would be diagnosed by cffDNA. Our center detected 4 diandric triploidies and 1 digynic triploidy. Therefore, based on the published DR of cffDNA testing by SNP, we conservatively predict that 4 of 5 triploidy would have been detected by cffDNA.
There were 20 unbalanced chromosomal rearrangements and a rare autosomal aneuploidy and 47 clinically significant CMA findings. We predict that these would not be detected by current cffDNA.
In total, we calculated that 99 of 220 chromosomal microscopic and submicroscopic changes (45%) would not be detected by cffDNA testing ( Table 2 ).
Fetal structural anomalies and increased NT/cystic hygroma were the indication for invasive testing in 42% of the cases (42/99) with chromosomal abnormalities assigned as nondetectable by cffDNA testing. Whereas in 21% of cases (21/99) with chromosomal abnormalities that were considered nondetectable by cffDNA, the only indication was AMA or maternal concern.
Of the 366 invasive procedures that were done for positive FTS, 9% procedures (n = 33) were screen positive. An abnormal karyotype other than the common chromosomal trisomies was present in 42% of the cases (14/33). Table 3 lists the indication for the invasive procedure in each category of chromosomal abnormality.
Common aneuploidy (n=97), n (%) | Monosomy X (n=21), n (%) | Other Sex chromosome trisomies (n=7), n (%) | Triploidy (n=5), n (%) | Mosaic (n=23), n (%) | Chromosomal microarray analysis (n=47), n (%) | Chromosomal rearrangement and a rare trisomy (n=20), n (%) | |
---|---|---|---|---|---|---|---|
Advanced maternal age | 12 (12.4) | 0 (0) | 1 (14.3) | 0 (0) | 6 (26.1) | 10 (21.3) | 1 (5) |
1st trimester screening (%) | |||||||
First trimester screening | 19 (19.6) | 0 (0) | 0 (0) | 0 (0) | 5 (21.8) | 8 (17) | 1 (5) |
Nuchal translucency/Cystic hygroma | 24 (24.7) | 16 (76.2) | 0 (0) | 0 (0) | 2 (8.7) | 3 (6.4) | 2 (10) |
2nd trimester screening | |||||||
Quadruple | 9 (9.3) | 0 (0) | 3 (42.9) | 0 (0) | 4 (17.4) | 5 (10.6) | 0 (0) |
Ultrasound scan | 27 (27.8) | 5 (23.8) | 0 (0) | 4 (80) | 3 (13) | 18 (38.3) | 10 (50) |
Elevated alpha fetoprotein | 0 (0) | 0 (0) | 1 (14.3) | 0 (0) | 1 (4.3) | 1 (2.1) | 0 (0) |
Other familial indications | |||||||
Parental chromosome rearrangement | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 4 (20) |
Previous affected pregnancy | 0 (0) | 0 (0) | 1 (14.3) | 0 (0) | 0 (0) | 1 (2.1) | 1 (5) |
Maternal concern/others | 0 (0) | 0 (0) | 0 (0) | 1 (20) | 1 (4.3) | 1 (2.1) | 1 (5) |
Positive cell-free fetal DNA | 6 (6.2) | 0 (0) | 1 (14.3) | 0 (0) | 1 (4.3) | 0 (0) | 0 (0) |
Estimation of number not detected by cell-free fetal DNA a | 7 (7.2) | 7 (33.3) | 1 (14.3) | 1 (20) | 14 (60.9) | 47 (100) | 20 (100) |